Predicting the location of northern goshawk nests: modeling the spatial dependency between nest locations and forest structure

Abstract

Northern goshawks interact with each other and their environment in a spatially dependent manner. However, finding the location of active goshawk nests (e.g. where eggs are laid) in a given year is difficult due to the secretive nature of the hawks in their forest environment, their annually variable attempts at nesting, and the extent of the area within a home range where they will nest. We used a Gibbsian pairwise potential model to describe the spatial dependency (1) among nest locations influenced by territoriality and (2) between nest locations and the environment for a large population of goshawks on the Kaibab National Forest’s (NNF) North Kaibab Ranger District (NKRD). Nest locations in a given year were regularly distributed at a minimum distance of 1.6 km between active nests; however, as the spatial scale increased (i.e. as distance between the nests increased), the degree of regularity decreased. Important forest predictors for nest locations included canopy closure, total basal area, proportion of basal area in ponderosa pine, spruce, fir, and aspen, maximum height of the understory vegetation, and presence/absence of seedlings and saplings. The probability of an occurrence of an active nest within a 10-m×10-m area was modeled using logistic regression. Spatial analysis, using nest spacing and habitat variables, indicated that potential active nest locations were abundant and randomly distributed throughout the NKRD. This supports the supposition that the availability of locations with high potential for nests is not limiting the goshawk population on the study area. Instead, territoriality, and what appear to be non-compressible territories, sets the upper limit to the nesting population. Ultimate choice of nest location was probably constrained by the availability of high potential locations within spaces defined by neighboring territories. Overall territory density, on the other hand, may reflect the abundance, quality, and accessibility of prey on the study area. This model can be used to evaluate the influence of forest management activities on the nesting goshawk population on the NKRD. The modeling technique described in this paper may be applied to other study areas, where vertebrate densities and the spatial resolution of habitat data may be less or greater than on this study, provided that new point process and pairwise potential models are developed for each area.